Several techniques are available for constraint satisfaction problems. Most of them lack real time capability and distributed computational properties. Configuration of complex systems has today become a competitive edge, e.g. within e-business. The other side of the same coin can be used for diagnosis and simulation of safe critical applications. This paper gives a scientific background and comparison of methods for constraint satisfaction problems implemented in real time environment. Especially, the technique and its features of the array-based logic will be illustrated by three examples.
The examples are:
1.Modelling and simulation of system wide protection applications of power systems
2.Modelling, simulation and implementation of safe critical switching in power systems
3.Modelling, simulation and implementation of safe critical systems of automatic train control
Array-based logic is a novel technology, which is founded on a geometrical representation of logic in terms of nested data arrays. In this representation, logical inferences are executed on finite domain systems by a few standardised array operations. In computational practice, array-based logic makes it possible to solve large-scale constraint problems with combinations very efficiently. The major advantages of the technology are completeness, compactness, and speed of simulation (real-time capability), all of which derive from intrinsic properties of the underlying geometrical model. A system in array-based logic is considered analogously to electrical networks and is thus described by the constraints of the isolated elements as well as the constraints of the interconnected elements (the system topology). All system constraints are eliminated in accordance with the methods known from electrical network theory and other engineering disciplines rooted in classical physics. The configuration space is represented explicitly in terms of nested data arrays and the system can therefore be simulated in real-time by co-ordinate indexing on these arrays. This data driven approach makes it suitable for parallel processing, which is an important quality for distributed systems.